Pseudopaline, a staphylopine-like metallophore involved in zinc and nickel uptake in Pseudomonas aeruginosa

Metal uptake is vital for all living organisms. In metal scarce conditions, a common bacterial strategy consists in the biosynthesis of metallophores, their export in the extracellular medium and the recovery of a metal-metallophore complex through dedicated membrane transporters. Staphylopine is a recently described metallophore distantly related to plant nicotianamine that contributes to the broad-spectrum metal uptake capabilities of Staphylococcus aureus. Here, we characterize a four genes operon (PA4837–PA4834) in Pseudomonas aeruginosa involved in the biosynthesis and trafficking of a staphylopine-like metallophore named pseudopaline. Pseudopaline differs from staphylopine with regard to the stereochemistry of its histidine moiety associated to an alpha ketoglutarate moiety instead of pyruvate. In vivo, the pseudopaline operon is regulated by zinc through the Zur repressor. The metal-uptake property of the pseudopaline system appears different from that of staphylopine with a predominant effect on nickel uptake, and on zinc uptake in metal scarce conditions mimicking a chelating environment, thus reconciling the regulation of the cnt operon by zinc with its function as a zinc importer under metal scarce conditions. AUTHOR SUMMARY Zinc is an essential micronutrients for bacteria, particularly important at the host-pathogen interface where the host tends to sequester metals in a so called nutritional immunity framework, and the pathogenic bacterium increases its metal uptake efforts in order to keep up with its metal requirements. Here we reveal a novel metallophore, named pseudopaline, which is synthesized and exported by Pseudomonas aeruginosa and serves for the uptake of nickel in metal poor media, and for the uptake of zinc in metal scarce conditions that mimic the chelating environment that presumably prevails within a host.

up with its metal requirements. Here we reveal a novel metallophore, named pseudopaline, 48 which is synthesized and exported by Pseudomonas aeruginosa and serves for the uptake of 49 nickel in metal poor media, and for the uptake of zinc in metal scarce conditions that mimic 50 the chelating environment that presumably prevails within a host. 51

INTRODUCTION 53
Divalent metals (Mn, Fe, Co, Ni, Cu and Zn) are essential micronutrients for all life forms, 54 and acquisition of these metals is therefore vital, particularly for bacterial pathogens in the 55 context of host-pathogen interactions. Indeed, there is a competition between the host, which 56 tends to sequester metals in a so called nutritional immunity framework, and the pathogenic 57 bacterium, which increases its metal uptake efforts in order to keep up with its metal 58 requirements (1, 2). Most pathogenic bacteria produce metallophores for metal uptake, with 59 siderophores being the most well-characterized metallophore family (3). Siderophores are 60 synthesized within the cell through non ribosomal peptide synthases (NRPS) or through a 61 NRPS independent system (NIS) and then are exported in the extracellular medium where 62 they scavenge iron. Extracellular iron siderophore complexes can be recognized and actively 63 transported into the periplasm by TonB dependent transporters (TBDT) in Gram-negative 64 bacteria, and usually ABC transporters in both Gram-negative and Gram-positive bacteria are 65 used for the crossing of the cytoplasmic membrane. There are many variations on this 66 common theme and, for example, some bacteria do not produce a specific type of siderophore 67 although they are able to use it for iron import (4). The siderophore pathway could also 68 prevent toxic accumulation of metals, which was particularly studied in the case of 69 Pseudomonas aeruginosa (5, 6). P. aeruginosa synthesizes two types of siderophores with 70 high iron affinity, pyochelin and pyoverdine, the latter being a demonstrated virulence factor 71 (7). 72 Metallophores specific for the uptake of metals other than iron have also been described, such 73 as the chalcophore methanobactin involved in copper uptake in methane-oxidizing bacteria (8, 74 9). Manganesophore have not been described as such, although TseM, a protein effector 75 secreted through a Type VI secretion system, was shown to play an important role in TBDT-76 dependent manganese uptake in Burkholderia thailandensis (10). There is also indirect 77 evidence for the existence of a nickelophore in Escherichia coli, although it has still to be 78 identified (11). Free histidine could also be used as a nickelophore in vivo for nickel uptake in 79 various bacteria (12, 13). Yersiniabactin, initially described as a siderophore, also exhibits 80 zincophore properties in Yersinia pestis (14, 15). Coelibactin, described in Streptomyces 81 coelicolor, may also represent a zincophore as it is synthesized by a NRPS under the control 82 of Zur, a zinc responsive repressor (16). 83 Staphylopine is a nicotianamine-like molecule that was recently described as a metallophore 84 with remarkable broad-spectrum specificity (17). In Staphylococcus aureus, staphylopine is 85 synthesized through the action of three soluble enzymes (SaCntKLM). SaCntK transforms L-histidine in D-histidine, SaCntL transfers an aminobutyrate moiety from S-87 adenosylmethionine (SAM) onto D-histidine, and SaCntM reductively condensates the 88 product of SaCntL (called xNA) with pyruvate. The staphylopine biosynthesis and trafficking 89 pathway is responsible for zinc, copper, nickel, cobalt and iron uptake, depending on the 90 growth conditions, and this system contributes to the virulence and fitness of S. aureus (17-91 19). The S. aureus cnt operon is partly conserved in P. aeruginosa, where homologues of the 92 cntL and cntM genes are found, albeit with 20-30% sequence identity at the protein level. 93 Upstream of cntL, a gene codes a predicted outer membrane protein belonging to the TBDT 94 family, and downstream of cntM, a gene codes for a predicted inner membrane protein 95 belonging to the EamA or DMT family (drug/metabolite transporter; Figure S1). 96 Transcriptomic approaches revealed that this gene cluster was highly expressed in a burn 97 wound model (20). This last gene was also identified as part of a novel siderophore pathway 98 that appeared vital for the growth of P. aeruginosa in airway mucus secretion (AMS) (21). 99 Finally, through a transcriptomic study of a Zur deficient strain, these four genes were found 100 in the top five regulated units, although most of them were annotated as hypothetical (22). 101 Here, we show that the four above-mentioned genes (here named cntO, cntL, cntM and cntI; 102 see supplementary table S1 for correspondence with locus tag in PAO1, PA7 and PA14 103 strains of P. aeruginosa) are part of an operon that is regulated by zinc level through the Zur 104 repressor. Using biochemistry and metabolomics approaches, we prove that the two 105 biosynthetic enzymes (PaCntL and PaCntM) synthesize a novel metallophore, which we 106 named pseudopaline, and which differs from staphylopine by the presence of a D-histidine 107 moiety instead of L-histidine, and an -ketoglutarate moiety instead of a pyruvate. A cntL 108 mutant strain is shown to be unable to synthesize pseudopaline and is impaired in its ability to 109 import nickel in a minimal media, supplemented or not with nickel. Under more stringent 110 conditions where a chelator such as EDTA is added to a minimal succinate (MS) medium, a 111 condition that presumably mimics the chelating environment prevailing within a host or in 112 AMS, we show evidence that the cntL mutant strain is unable to import zinc, therefore 113 reconciling the regulation of this operon by zinc with its function as a zinc importer 114 functioning in metal scarce conditions. 115

RESULTS AND DISCUSSION 117
The cnt operon of P. aeruginosa is regulated by zinc level through the zinc-responsive 118

regulator Zur 119
In silico analysis of the cnt gene cluster of P. aeruginosa PA14 strain indicated two 120 overlapping open reading frames between cntL and cntM and between cntM and cntI, 121 classically observed in operonic structures ( Figure S1). Further screening of the upstream cnt 122 sequence for promoter regions using Bprom software (23), revealed a σ70 promoter in the 200 123 base-pairs upstream from the annotated cntO ATG codon ( Figure S1). Interestingly, a putative 124 Zur binding box "GTTATagtATAtC" can be identified overlapping the -10 box of the 125 predicted σ70 promoter, (22,24). This in silico analysis supports an operonic organization of 126 the four cnt genes and strongly suggests a transcriptional activation of this operon under zinc 127 depletion through the Zur repressor (25,26). In order to test this hypothesis, we performed 128 RT-PCR experiments using as templates RNA and cDNA generated from a WT PA14 strain 129 grown in minimal succinate (MS) medium known to contains low levels of metals, including 130 zinc (5). The successful amplification of the four cnt gene transcripts ( Figure S1) indeed 131 indicated their induction when cells were grown in a MS medium. The specific amplification 132 of the three cnt intergenic regions confirmed that the four cnt genes are co-transcribed in one 133 single transcript and therefore constitute an operon. 134 To validate at the protein level the transcriptional regulation of the cnt genes, we followed by 135 immunoblotting the PaCntL production under various growth conditions. In this respect, we 136 constructed a cntL mutant strain producing a chromosomally encoded V5-tagged PaCntL 137 (∆cntL::cntL V5 ). In this strain, the recombinant cntL V5 gene was placed under the predicted cnt 138 promoter region and inserted at the att site of the P. aeruginosa genome. In agreement with 139 our transcriptional data, immunoblotting experiments indicated that, the recombinant 140 PaCntL V5 is only produced in MS medium and not in a rich medium such as the LB medium 141 ( Figure 1A). Presumably, this is due to the low metal content of the MS medium as compared 142 to the LB medium. We then tested whether the cntL transcription was subject to metal 143 repression by checking PaCntL V5 production in MS medium supplemented with various 144 concentrations of the most representative metals. Dot-blot experiments showed a specific loss 145 of PaCntL V5 production in MS medium supplemented with as low as 0.1 μM of ZnSO 4 . An 146 addition of iron, nickel or cobalt at concentrations equivalent or above the one found in LB 147 rich medium (5) has no negative effect on PaCntL V5 production ( Figure 1B). The hypothesis 148 of a Zur repressor regulating the cnt operon was then tested through the construction of a its repressor activity ( Figure 1C). Taken together, these data therefore demonstrate that the cnt 152 operon of P. aeruginosa is negatively regulated by zinc, most probably through the binding of 153 a Zn-Zur repressor complex onto the predicted Zur binding motif identified in the σ70 154 promoter, thus preventing the recruitment of RNA-polymerase.  The levels of all the other metals were not changed in the ΔcntL mutant strain compared to the 233 WT strain (data not shown). A similar 90% reduction in intracellular nickel concentration was 234 also observed when the culture was supplemented with 1μM NiCl 2 ( Figure S4), thus 235 confirming that nickel uptake was predominantly performed by pseudopaline in these metal-236 poor media. We were intrigued by the apparent contradiction between the clear cnt operon 237 regulation by zinc, and the absence of any effect on zinc uptake. A possible explanation is that 238 the effect of cnt could be masked by the effect of a zinc ion importer such as the ZnuABC 239 zinc transport system described in P. aeruginosa (22). In an attempt to discriminate between 240 both transport systems, we sequestered free metal ions by supplementing the growth medium 241 with increasing concentrations of EDTA, a chelating agent for divalent metals. Interestingly, 242 although we did not observe any effect using 10 M EDTA, the supplementation with 100M 243 EDTA ultimately revealed a pseudopaline-dependent zinc uptake, with a 60% decrease of 244 intracellular zinc content in the ΔcntL mutant strain in comparison with the WT strain ( Figure  245 4B). The complemented strain accumulated zinc to a level comparable to the WT. In these 246 chelating conditions the pseudopaline-dependent nickel import is abolished ( Figure 4A), 247 hence proving a direct link between pseudopaline and zinc uptake in metal scarce conditions 248 with competing zinc chelators. These conditions may prevail at the host-pathogen interface 249 where metal binding proteins such as calprotectin are produced by the host (29, 30), or in 250 AMS where metals are complexed in a nutritional immunity framework (1,21). 251

Model of pseudopaline synthesis and transport pathway in P. aeruginosa 253
We next investigated the putative roles of the two membrane proteins that are found in the cnt 254 operon of P. aeruginosa by determining the pseudopaline level in the extracellular and 255 intracellular fractions of WT and mutant strains ( Figure 5A and 5B, respectively). With 256 regard to PaCntO, we found a small decrease in the extracellular content of pseudopaline in 257 the cntO mutant strain in comparison with the WT strain. However, we also found that this 258 cntO mutant strain was partly impaired in nickel accumulation ( Figure S5). Altogether, and 259 because PaCntO belongs to the TBDT family of extracellular transporter, its most probable 260 role is in the import of pseudopaline-metal complexes, although it is not excluded that other 261 proteins of this family could participate in this process. Next, we noted a large decrease in the 262 extracellular pseudopaline level in the cntI mutant strain in comparison with the WT strain, 263 with a concomitant increase in the intracellular space, consistent with a role of PaCntI in 264 pseudopaline export. It is also interesting to note that a cntI mutant strain is virtually unable 265 to grow in AMS (21). The most probable scenario is that this mutant is deficient in metal 266 content, including zinc, but pseudopaline accumulation in the cytoplasmic space actually 267 worsens the situation by chelating an already poorly available metal. This assumption is 268 supported by our finding that a double cntLcntI mutant supresses the detrimental growth 269 defect of the single cntI mutant strain, ie the absence of pseudopaline restores the normal 270 growth of a mutant devoid of the pseudopaline exporter ( Figure S6). A model recapitulating 271 the pseudopaline pathway is shown in Figure 5. 272 It is interesting to note the differences and similarities between staphylopine and pseudopaline 273 and between their respective biosynthetic pathways ( Figure S7). On one hand, pseudopaline 274 differs from staphylopine by the incorporation of a L-histidine instead of a D-histidine moiety 275 in staphylopine, thus explaining the absence of amino acid racemase in P. aeruginosa. 276 Another particularity of pseudopaline is the use of an KG moiety instead of pyruvate as 277 substrate for the second reaction mediated by PaCntM. Together this leads to two species-278 specific metallophores that might give a selective advantage in a competing environment. The 279 fact that staphylopine and pseudopaline belong to Gram-positive and Gram-negative bacteria 280 has important consequences on their respective transport mechanisms across the two types of 281 bacterial envelopes. Although the transporters of staphylopine are well identified, the outer 282 membrane exporter pseudopaline and inner membrane importer of the pseudopaline-metal 283 complex remains to be discovered ( Figure 5). Recycling of the metallophore could also take 284 place in P. aeruginosa, as recently exemplified in the case for pyoverdine (31). An interesting aspect of this work is the discovery of two different pathways for the export of these 286 nicotianamine-like bacterial metallophores. Whereas S. aureus uses a protein belonging to the 287 MFS family (SaCntE) for staphylopine export, P. aeruginosa uses a protein belonging to the 288 DMT family of transporters, with PaCntI possessing two predicted EamA domains for 289 pseudopaline export. In the view of their importance in the growth or virulence of these 290 important human pathogens (19, 21), they could emerge as attractive targets for novel 291 antibiotic development.